Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A 3D reconstruction apparatus including material appearance modeling, comprising: a light source arc including a plurality of light sources configured to irradiate an object located at a photographing stage; a camera arc including a plurality of cameras configured to photograph the object located at the photographing stage; a driving operation unit configured to rotate the light source arc and the camera arc; the photographing stage on which the object is disposed; and a control unit configured to control the operations of the plurality of light sources included in the light source arc, the operations of the plurality of cameras included in the camera arc, and the driving operation unit, wherein the control unit is further configured to: rotate the light source arc and the camera arc at a constant interval based on an operation axis through the driving operation unit in order to acquire a plurality of image samples of the object, extract an image region from each of the image samples which is necessary for 3D reconstruction and reflection characteristic estimation of the object, compress the plurality of image samples based on the extracted image region by merging the extracted image region in one file, estimate a lookup table for entire vertices based on data interpolation through reflection characteristic data of each point stored in the merged file, estimate basis of the lookup table for entire vertices to compress the lookup table for entire vertices, and reconstruct a 3D model corresponding to the object with 3D shape and surface textures using the estimated basis, wherein the surface textures include color and reflection corresponding to the object, wherein the plurality of image samples are acquired by operating the light source and the camera at a predefined location in a hemi-sphere direction with respect to the object.
This invention relates to a 3D reconstruction apparatus that captures and processes images of an object to generate a detailed 3D model with accurate surface textures, including color and reflection characteristics. The apparatus includes a light source arc with multiple light sources and a camera arc with multiple cameras, both arranged around a photographing stage where the object is placed. A driving unit rotates the arcs at fixed intervals around an operation axis to capture multiple image samples from different angles, simulating a hemispherical light and camera arrangement. A control unit manages the light sources, cameras, and rotation, ensuring synchronized operation. The system extracts relevant image regions from each sample, compresses them into a single file, and uses reflection data from these regions to estimate a lookup table for all vertices of the object. Data interpolation refines this table, and basis estimation further compresses it. The final 3D model integrates shape and surface textures, including color and reflection properties, derived from the processed image data. The method ensures high-fidelity reconstruction by leveraging structured multi-angle imaging and efficient data compression techniques.
2. The 3D reconstruction apparatus according to claim 1 , wherein the control unit is configured to: extract feature points of each of the image samples by using the plurality of image samples, match the feature points to calculate poses of the cameras with respect to directions and rotations of the cameras with respect to each of the image samples, generate 3D coordinates in a virtual space of the feature points when the poses of the cameras are calculated, and generate a 3D point cloud.
A 3D reconstruction apparatus is designed to create detailed three-dimensional models from multiple two-dimensional image samples captured by cameras. The apparatus addresses the challenge of accurately reconstructing spatial information from 2D images, which is essential for applications in robotics, autonomous navigation, and virtual reality. The system includes a control unit that processes the image samples to extract feature points, which are distinctive elements within each image. These feature points are then matched across the image samples to determine the relative poses of the cameras, including their directions and rotations. By calculating these poses, the system generates 3D coordinates for the feature points in a virtual space, effectively mapping their positions in three dimensions. The result is a 3D point cloud, a collection of data points that represent the surface geometry of the objects or scenes captured in the images. This point cloud can be used to construct a detailed 3D model, enabling applications such as object recognition, environmental mapping, and virtual simulations. The apparatus leverages advanced image processing techniques to ensure high accuracy in pose estimation and 3D coordinate generation, making it suitable for environments where precise spatial reconstruction is required.
3. The 3D reconstruction apparatus according to claim 2 , wherein the control unit is configured to generate a 3D reconstruction result in a mesh shape by using a Poisson surface reconstruction method.
A 3D reconstruction apparatus generates a three-dimensional model of an object from captured images or sensor data. The apparatus includes a control unit that processes the input data to reconstruct the object's surface geometry. Specifically, the control unit applies a Poisson surface reconstruction method to generate a mesh-shaped 3D reconstruction result. This method mathematically estimates the surface of the object by solving a Poisson equation derived from the input data, producing a smooth and accurate mesh representation. The apparatus may also include an imaging unit to capture the necessary data, such as images or depth measurements, and a storage unit to retain the reconstruction results. The Poisson surface reconstruction method is particularly effective for handling noisy or sparse input data, ensuring high-quality surface reconstruction. The apparatus is useful in applications like medical imaging, industrial inspection, and virtual reality, where precise 3D models are required.
4. The 3D reconstruction apparatus according to claim 1 , wherein an angle between the light sources disposed in the light source arc is constant.
A 3D reconstruction apparatus is designed to capture and process depth information from an object using structured light projection. The apparatus includes multiple light sources arranged in an arc-shaped configuration to illuminate the object from various angles. The key feature is that the angle between adjacent light sources in the arc is constant, ensuring uniform angular spacing. This uniform spacing improves the accuracy and consistency of the depth data captured by the system. The apparatus may also include a camera or sensor to capture reflections of the projected light patterns, which are then processed to generate a 3D model of the object. The constant angular spacing of the light sources helps minimize errors in depth estimation by providing a predictable and evenly distributed set of illumination angles. This design is particularly useful in applications requiring high-precision 3D reconstruction, such as industrial inspection, medical imaging, or augmented reality. The apparatus may further incorporate calibration techniques to enhance the accuracy of the depth measurements.
5. The 3D reconstruction apparatus according to claim 1 , wherein an angle between the cameras disposed in the camera arc is constant.
A 3D reconstruction apparatus includes multiple cameras arranged in a curved arc to capture images of an object from different angles. The apparatus ensures that the angle between adjacent cameras in the arc is constant, providing uniform angular spacing for consistent image capture. This setup improves the accuracy and reliability of the 3D reconstruction process by minimizing gaps or overlaps in the captured data. The apparatus may also include a light source to illuminate the object, enhancing image quality and reducing shadows. The cameras are synchronized to capture images simultaneously, ensuring temporal consistency in the reconstruction. The apparatus may further include a processing unit that combines the captured images to generate a detailed 3D model of the object. This method is particularly useful in applications requiring high-precision 3D modeling, such as industrial inspection, medical imaging, or virtual reality content creation. The constant angular spacing between cameras ensures balanced coverage, reducing errors in depth perception and surface reconstruction. The apparatus may also incorporate calibration mechanisms to maintain optimal camera alignment and focus, further improving reconstruction accuracy.
6. A method for controlling a 3D reconstruction apparatus including material appearance modeling, the method comprising: controlling, via a control unit of the 3D reconstruction apparatus, a light source arc and a plurality of light sources included in the light source to irradiate light on an object disposed on a photographing stage; and controlling, via the control unit, a camera arc and a plurality of cameras included in the camera arc to photograph the object disposed on the photographing stage, wherein controlling the light source arc and controlling the camera arc comprises rotating the light source arc and the camera arc at a constant interval based on an operation axis in order to acquire a plurality of image samples of the object, the method further comprises: extracting, via the control unit, an image region from each of the image samples which is necessary for 3D reconstruction and reflection characteristic estimation of the object, compressing, via the control unit, the plurality of image samples based on the extracted image region by merging the extracted image region in one file, estimating, via the control unit, a lookup table for entire vertices based on data interpolation through reflection characteristic data of each point stored in the merged file, estimating, via the control unit, basis of the lookup table for entire vertices to compress the lookup table for entire vertices, and reconstructing, via the control unit, a 3D model corresponding to the object with 3D shape and surface textures using the estimated basis, wherein the surface textures include color and reflection corresponding to the object, wherein the plurality of image samples are acquired by operating the light source and the camera at a predefined location in a hemi-sphere direction with respect to the object.
This invention relates to a method for controlling a 3D reconstruction apparatus that models material appearance, addressing challenges in capturing and reconstructing accurate 3D models with realistic surface textures, including color and reflection properties. The method involves a light source arc and a camera arc, each equipped with multiple light sources and cameras, respectively. The light source arc and camera arc rotate at constant intervals around an object placed on a photographing stage to capture multiple image samples from predefined positions in a hemispherical direction relative to the object. A control unit processes these images by extracting necessary regions for 3D reconstruction and reflection characteristic estimation, then compresses the image data by merging these regions into a single file. The control unit then estimates a lookup table for all vertices of the 3D model using interpolation based on reflection characteristic data from the merged file. To optimize storage and processing, the lookup table is compressed by estimating its basis. Finally, the control unit reconstructs a 3D model of the object, incorporating both its 3D shape and surface textures, which include color and reflection properties derived from the captured images. This approach ensures high-fidelity 3D reconstruction with accurate material appearance modeling.
7. The method according to claim 6 , further comprising: extracting, via the control unit, feature points of each of the image samples by using the plurality of image samples; matching, via the control unit, the feature points to calculate poses of the cameras with respect to directions and rotations of the cameras with respect to each of the image samples; generating, via the control unit, 3D coordinates in a virtual space of the feature points when the poses of the cameras are calculated; and generating, via the control unit, a 3D point cloud.
This invention relates to computer vision and 3D reconstruction, specifically addressing the challenge of accurately determining camera poses and generating 3D point clouds from multiple image samples. The method involves extracting feature points from each image sample to identify distinctive visual markers. These feature points are then matched across the images to calculate the poses of the cameras, including their directions and rotations relative to each image. Once the camera poses are determined, the system generates 3D coordinates in a virtual space for the matched feature points. These coordinates are used to construct a 3D point cloud, which represents the spatial arrangement of the feature points in three dimensions. The process leverages multiple image samples to enhance accuracy and robustness in reconstructing the 3D environment. This approach is particularly useful in applications such as augmented reality, robotics, and autonomous navigation, where precise spatial mapping is essential. The method ensures that the generated 3D point cloud accurately reflects the real-world positions of the feature points, enabling reliable spatial analysis and interaction.
8. The method according to claim 7 , further comprising generating, via the control unit, a 3D reconstruction result in a mesh shape by using a Poisson surface reconstruction method.
This invention relates to 3D reconstruction techniques, specifically improving the accuracy and efficiency of generating 3D models from input data. The method addresses challenges in creating high-quality 3D reconstructions, particularly when dealing with noisy or sparse input data, by employing advanced surface reconstruction algorithms. The process involves capturing input data, such as point clouds or depth maps, and processing this data to generate a 3D reconstruction. A control unit manages the reconstruction process, applying filtering and noise reduction techniques to enhance data quality before reconstruction. The method further includes generating a 3D reconstruction result in a mesh shape using a Poisson surface reconstruction method. This approach optimizes surface smoothness and detail preservation, ensuring accurate and visually coherent 3D models. The Poisson surface reconstruction method is particularly effective for handling incomplete or irregular input data, as it interpolates surfaces based on gradient constraints, resulting in smooth and continuous meshes. The control unit ensures that the reconstruction parameters are dynamically adjusted to adapt to varying input data characteristics, improving robustness and adaptability. By integrating these techniques, the invention provides a reliable and efficient solution for generating high-fidelity 3D reconstructions from diverse input sources, suitable for applications in computer vision, robotics, and 3D modeling.
Unknown
March 10, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.